Tudor Cebere

I am a third-year PhD student at Inria 🇫🇷, fortunate to be supervised by Aurélien Bellet. I am a recipient of the Google PhD Fellowship 2025 in Privacy, Safety, and Security. Since August 2024, I am also an Associate in Computer Science at Harvard, working with the OpenDP group. During my PhD, I have also completed research visits at the Vector Institute 🇨🇦 (hosted by Nicolas Papernot) and Harvard 🇺🇸 (hosted by the OpenDP group).

My research focuses on differential privacy and its role in building trustworthy machine learning systems. I study the design and analysis of private algorithms, with a particular interest in how their theoretical guarantees translate to practical deployments. A major direction of my work is privacy auditing: developing practical and statistically principled methods to test, stress, and better understand the privacy guarantees provided by algorithms and their implementations. More broadly, I aim to bridge the gap between formal worst-case privacy guarantees and the guarantees that can be meaningfully assessed in realistic settings, making privacy claims more transparent, measurable, and reliable for practitioners.

I graduated from École normale supérieure de Lyon (ENS Lyon) with a masters in Theoretical CS, where I was fortunate to receive an Ampère Scholarship of Excellence. Before that, I received my BSc in EECS from Politehnica University of Bucharest 🇷🇴.

📄 You can find my latest CV here.

Publications

2026	9. Lumberjack: Better Differentially Private Random Forests through Heavy Hitter Detection in Trees – Christian Janos Lebeda, David Erb, Tudor Cebere, Aurélien Bellet – arXiv – Preprint – Code
	8. Privacy Auditing with Zero (0) Training Run – Tudor Cebere, Mathieu Even, Linus Bleistein, Aurélien Bellet – arXiv – Preprint
	7. Privacy in Theory, Bugs in Practice: Grey-Box Auditing of Differential Privacy Libraries – Tudor Cebere, David Erb, Damien Desfontaines, Aurélien Bellet, Jack Fitzsimons – The 26th Privacy Enhancing Technologies Symposium – Conference – Code
	6. Membership Inference Attacks from Causal Principles – Mathieu Even, Clément Berenfeld, Linus Bleistein, Tudor Cebere, Julie Josse, Aurélien Bellet – arXiv – Preprint
2025	5. Private Rate-Constrained Optimization with Applications to Fair Learning – Mohammad Yaghini, Tudor Cebere, Michael Menart, Aurélien Bellet, Nicolas Papernot – Proceedings of the 14th International Conference on Learning Representations. – Conference – Code
2024	4. Tighter Privacy Auditing of DP-SGD in the Hidden State Threat Model – Tudor Cebere, Aurélien Bellet, Nicolas Papernot – Proceedings of the 13th International Conference on Learning Representations. – Conference
2024	3. Confidential-DPproof: Confidential Proof of Differentially Private Training – Ali Shahin Shamsabadi, Gefei Tan, Tudor Cebere, Aurélien Bellet, Hamed Haddadi, Nicolas Papernot, Xiao Wang, Adrian Weller – Proceedings of the 12th International Conference on Learning Representations. – Conference (+spotlight)
2021	2. Syft 0.5: A platform for universally deployable structured transparency – Adam James Hall, Madhava Jay, Tudor Cebere, Bogdan Cebere et al. – Distributed and Private Machine Learning Workshop at the 9th International Conference on Learning Representations. – Workshop
2021	1. Pyvertical: A vertical federated learning framework for multi-headed splitnn – Daniele Romanini, Adam James Hall, Pavlos Papadopoulos, Tom Titcombe, Abbas Ismail, Tudor Cebere, Robert Sandmann, Robin Roehm, Michael A Hoeh – Distributed and Private Machine Learning Workshop at the 9th International Conference on Learning Representations. – Workshop